reference/generated/torchrl.record.loggers.monitoring.LoggerMonitor

Run in Google Colab

Colab

Download Notebook

Notebook

View on GitHub

GitHub

LoggerMonitor

torchrl.record.loggers.monitoring.LoggerMonitor(logger: Logger, *, poll_interval: float = 1.0, background: bool = True, on_error: Literal['warn', 'raise', 'ignore'] = 'warn') → None

A pull-based monitor logging operational statistics of collectors and replay buffers.

The monitor periodically requests a cheap stats() snapshot from each watched object, decides whether the snapshot is due for logging according to the per-target schedule, derives rates from cumulative counter deltas, namespaces the metrics as "<name>/<metric>" and forwards them to a single Logger.

Because snapshots are pulled from the monitor’s own thread (or from explicit step() calls), no logging work is ever executed on the collection, write or sampling hot paths of the watched objects, and a slow logging backend can only delay the next poll, never build an unbounded backlog. The monitor works with any object exposing a stats() method returning a flat mapping of scalars: in particular local, multiprocessing and Ray collectors and replay buffers share the same interface.

The monitor never takes ownership of the logger or of the watched objects: stopping the monitor leaves both running, and shutting down the logger remains the caller’s responsibility.

Parameters:

logger (Logger) – the output sink. Any TorchRL logger works, including loggers running as a service (service_backend="ray").

Keyword Arguments:

• poll_interval (float, optional) – how often the background thread polls the watched objects, in seconds. Counter schedules are evaluated at each poll, so this bounds their resolution. Defaults to 1.0.

• background (bool, optional) – if True (default), entering the monitor context (or calling start()) spawns a daemon thread that polls every poll_interval seconds. If False, nothing runs in the background and the user drives the monitor through explicit step() calls, which keeps tests and deterministic loops reproducible.

• on_error (str, optional) – behavior when polling or logging a target fails: "warn" (default) logs a warning once per failure streak through the torchrl logger, "ignore" silently skips, and "raise" propagates the exception (in background mode this terminates the monitor thread).

Examples

>>> import tempfile
>>> import torch
>>> from torchrl.data import LazyTensorStorage, ReplayBuffer
>>> from torchrl.record import CSVLogger
>>> from torchrl.record.loggers.monitoring import Every, LoggerMonitor
>>> logger = CSVLogger(exp_name="monitor_demo", log_dir=tempfile.mkdtemp())
>>> rb = ReplayBuffer(storage=LazyTensorStorage(100))
>>> monitor = LoggerMonitor(logger, background=False)
>>> handle = monitor.watch(rb, name="replay_buffer", step="write_count")
>>> _ = rb.extend(torch.arange(10))
>>> logged = monitor.step()
>>> print(logged["replay_buffer"]["replay_buffer/size"])
10.0

The typical background usage mirrors the RFC acceptance example:

>>> with LoggerMonitor(logger) as monitor:  
...     monitor.watch(collector, name="collector", schedule=Every.counter("frames", 10_000))
...     monitor.watch(replay_buffer, name="replay_buffer", schedule=Every.seconds(5), step="write_count")
...     collector.start()
...     run_training()